Extended Abstract
 
Introduction
To analyze complex ecological data, it is necessary to employ flexible and robust analytical methods that can handle non-linear relationships, interactions, and missing data control. The generalized additive model is a simple method for investigating species' reactions to environmental variables, and the results can be easily interpreted. Multivariable regressions such as the Monotonic increase model can play a role in expressing the ecological niche of a particular species. In this ecological domain, non-living and living factors have mutual effects, but the relative importance of living factors, such as species competition, is unclear when compared to non-living factors. More research is necessary for this issue. The response curve studies only investigate the behavior of the species along an environmental gradient, if multiple factors come together to determine the distribution and behavior of the species, it is important to take this into account. The comparison between the findings of researchers in different regions on the species Astragalus curvirostris Boiss consistently shows the extreme effectiveness of the species in terms of environmental factors or their combination. Understanding the relationship between plants and environmental factors and how plants respond to changes in environmental factors is one of the important topics in plant ecology. Unfortunately, in recent years, disturbances such as livestock grazing, changes in land use, and climate change have caused the destruction of A. curvirostris habitats. Many pastures and forests throughout Iran experience vegetation destruction, loss of biodiversity, and soil erosion due to these threats. To restore degraded pastures in steppe and semi-steppe areas, it is necessary to understand A. curvirostris' response to environmental variables and model its distribution. Comprehending the positive and negative effects of environmental factors, as well as the factors that inhibit and support the growth, establishment, and distribution of Astragalus curvirostris Boiss., it became possible to use it for the restoration and modification of appropriate and similar pasture ecosystems.
 
Material and Methods
The habitat of A. curvirostris was selected in a way that includes a wide range of abundance of the studied species. Therefore, according to the 40-hectare enclosure range in the research site of Zaghe rangelands plants in Lorestan province, this habitat was selected as a study site. Then, according to the gradient of environmental changes, five transects with a length of 400 meters and distances of 100 meters were used. Each of them had six plots with dimensions of 4 meters by 4 meters set up according to the species' sizes, with equal distances. The ecological unit consisted of 30 plots and recorded the geographic location, quantitative and qualitative vegetation amounts in each plot. Additionally, the Lorestan Meteorological Department provided climatic parameters. In order to investigate the effect of environmental factors on species distribution, a soil sample with three replicates (0-30 cm) was taken from each of the plots and their physical and chemical characteristics including soil texture, acidity, percentage of neutralized substances, absorbable phosphorus, absorbable potassium, organic carbon and total nitrogen were measured. Considering the importance of species in fodder production and soil protection in pastures, the canonical correspondence analysis was used to determine the factors affecting changes in the species composition. To investigate the changes in the performance of this species along the slope of the environmental factors from the Generalized Additive Models was used. A random-systematic method was used to sample environmental and plant characteristics from 2016 to 2018.
 
Results and Discussion
The forward selection method in conic ranking was used to select 4 variables from 19 primary variables while investigating the effect of a set of environmental factors on vegetation changes in communities. The geographical direction of the range was determined by the percentage of organic carbon, clay, acidity, and litter on the soil surface, as well as the topographic group. Investigating the correlation between the percentage of A. curvirostris vegetation and the studied ecological factors showed that distribution of this species has positive correlation with factors such as the percentage of O.C and N. Applying the GAM with Poisson error distribution showed that the variables, height above sea level, percentage of organic matter and soil nitrogen, as well as the percentage of stone are effective on the yield of the species. Investigating the performance of the species in relation to the variable of height above sea level, and the percentage of stones and pebbles from the Monotonic decrease model and vice versa, the response of this species along the slope of changes in the amount of organic carbon, soil nitrogen and the percentage of soil litter from the Monotonic increase model mathematical expression of relationships between environmental variables and biological and biophysical characteristics is only an aid for interpreting field observations. The ecological needs of the species are depicted in this model, which can be useful for natural resource managers in improving pastures in similar areas. Corrective and pasture management programs can use forecasting models that have adequate accuracy to suggest species that are compatible with the region's conditions. Therefore, it is important to include disturbances as predictors in regression-based distribution modeling. In summary, the dynamic of biological factors in grassland ecosystems makes it impossible to definitively assume even the strongest correlations, both in static and dynamic studies, in grassland ecosystems. This information, along with the field investigation, provides the appropriate information.